The current study investigated psychometric properties of the Family Affective Attitude Rating Scale (FAARS) for assessing parents' thoughts and feelings about their child, coded from a 5-min speech sample. Parental affective attitudes derive from previous experiences of parenting and child behavior, representations of the parent-child relationship and broader parental characteristics. Data were collected from mother-child dyads at ages 2 and 3 (N = 731; 49% female) from a multi-ethnic and high-risk community sample. Multi-informant observations of parenting and questionnaire measures were used to test construct and discriminant validity. FAARS showed good internal consistency and high inter-rater agreement. Affective attitudes were related to mothers' perceptions of their daily hassles, their reports of conflict with their child, and observed measures of positive and harsh parenting. Negative affective attitudes uniquely predicted later child problem behavior, over and above maternal reports of and observed measures of parenting. Overall, results support the validity of FAARS coding in mothers of preschoolers, a previously untested group. FAARS is a novel measure, directly assessing maternal perceptions of the parent-child relationship, and indirectly providing an index of maternal affect, stress, and depressive symptoms. Its brevity and cost-effectiveness further enhance the potential use of the FAARS measure for clinical and research settings.